Certain error correction techniques use “soft” bits to improve the error correction capability. For example, soft bits are generally redundant forms of hard bits. A soft bit may indicate a reliability of a hard bit read from a storage element. With a flash memory device, a soft bit may be used to indicate reliability of a “hard” bit based on the relative distance between soft read voltage and a threshold read voltage. In whatever form, soft bits may improve decoding by indicating which bits of data are likely to be “incorrect” such that they can be flipped prior to error correction.